Research On Predictive Control Of Grate Cooler Based On Typical Working Condition Model

The grate cooler is the key equipment for the clinker cooling system.The cooling effect is not only related to the quality of cement clinker,but also has an indirect influence on other aspects of cement production.At present,the control of grate cooler in our country is mainly manual control and semi-automatic control.The subjectivity of manual control is strong and the randomness is large.There are many problems such as nonlinear,large delay and frequent fluctuation of working conditions in grate cooler.The research on the grate cooler is changing from the traditional control algorithm to the advanced intelligent control algorithm.Therefore,it is necessary to study the predictive control of grate cooler.Based on understanding of the process of the cement grate cooler,this paper combines with the operation experience of the operator and takestypical working conditions as the breakthrough point.This paper adopts the dynamic matrix control algorithm and presents a predictive control scheme for grate cooler based on typical working condition.At last,we develop the predictive control software of the grate cooler by using modularization.After debugging,the system is applied to the practical control.The application result shows that the control software has a good control effectand has strong practicability.The detailed work of this paper is as follows:(1)The typical working condition template of grate cooleris established.Aiming at the problems such as nonlinear and frequent fluctuation of the working conditions in the grate cooler,we take K-means clustering method to cluster the historical data and thenthree typical conditions are divided.The reference values of other parameters are obtained according to the dominant variables(grate pressure)and the typical working condition accounts for about 90 percent of the whole grate cooler working condition.The rationality and correctness of the method are verified by practical application.In this paper,the grate cooler is studied under the typical working condition,which is the basis for the next research.(2)Study on prediction model of grate cooler under typical working condition.According to the typical condition,weselect typical historical data andthen considering the numerous field interference,the data will be error inevitably.Thus,the mean filtering method is used to filter the typical historical data and then the least square method is used to build the prediction model of grate cooler.Because of the switching of working condition in the operation of the grate cooler,it is bound to switch the model when the working conditions change.In order to ensure the stability of the switching process,the fuzzy weighted method is used to switch the model in different conditions.(3)Research onDMCprediction controller of grate cooler under typical working condition.On the basis of the prediction model,the DMC controller is designed and the optimal control parameter of each working condition is established by MATLAB simulation.Considering the practical application,for the issue of controller switching,the fuzzy weighted method is used to realize the smooth switching and the stability of the controller is analyzed.(4)The Engineering implementation of predictive control software for grate cooler.According to the above research contents,the predictive control software of grate cooler is developed by using modularization idea.The software is composed of the following parts: the data acquisition system,the database storage system and control algorithm system.In the actual production site,the software is installed in theengineer station by the way of plug-in package.Through the actual operation of the site,the software achieved a good control effect and had a strong practicality.At the same time,the significance of this research is verified.